Two hundred past customers in a spreadsheet: the maintenance nobody calls about

You fitted the air conditioning two years ago. It works, the customer was happy, and you have not spoken since.

Two seasons have passed. The unit should be serviced annually, and the customer either does not know that or knows and has not thought about it. Early in the summer, when it stops cooling properly, they search online — and there is no guarantee they find you.

Their name is among your invoices. So is the address, and which unit is installed. It simply never turned into an enquiry.

Why this is hard

In principle it is a list. In practice it isn't.

The data is not in one place. Partly in the invoicing system, partly in notes on your phone, partly only in your head.

Timing is not neutral. Air conditioning cleaning is worth offering in April, not in July when everyone wants it at once. Boiler checks belong to the run-up to the heating season.

Not everyone needs the same thing. One address has two units, one customer called last year themselves, and at one you replaced the unit last year — that one does not need the same offer now.

Someone who declined once should not be pestered again, but you do not want to write them off permanently either.

And the decisive point: this task never has a deadline. It can always wait until tomorrow, because nobody will ask about it today.

What the system does

It gathers what is already in your data: which unit is at which address, when it was installed, when it was last serviced.

From that it builds a schedule of who is worth contacting and when. Not everyone at once, but paced ahead of the season, at a volume your calendar can absorb.

It drafts the message too. Not a circular: addressed to that customer, referring to their unit, with the date you were last there.

You approve it before it goes out. Not because the system is unreliable, but because you know who you do not want to contact — and it does not know that, nor does it need to.

Whoever replies becomes an appointment. Whoever does not gets one scheduled reminder, and then drops out of the queue.

How it decides

Three things set the order.

The first is the cycle. If a unit needs annual servicing and the last visit was fourteen months ago, it is due.

The second is the season. Spring months for air conditioning, the period before the heating season for gas appliances. The system does not time to the calendar anniversary, but to when it makes sense to the customer.

The third is history. Someone who responded last year is contacted earlier. Someone who ignored two approaches moves down the list or off it.

Here less is more. A message sent at the wrong moment is not neutral: it can lose you the customer who would otherwise have come back. So the system would rather skip someone than write to them on a guess.

Edge cases

The customer has moved. Same address, different person. That surfaces in the reply, and the system stops contacting them.

They had it done by someone else in the meantime. It happens. That is no reason to delete them — in two years it comes round again.

Two units at one address, installed at different times. One message should go, not two.

Warranty obligations. Where the manufacturer's warranty depends on annual servicing, the message is not an offer but a warning — and it has to be worded differently.

Statutory obligations. For gas appliances, periodic inspection is not a choice. There the message serves the customer's own interest, and it is fine to say so plainly.

Which mistake actually costs money

Not the message you did not send. The customer you lost.

Someone who has not heard from you in two years will not call you next time. Not because they were unhappy, but because you did not come to mind. The next job, the replacement, the referral — all of it goes to whoever was reachable at that moment.

This is the most expensive kind of loss, because the acquisition already happened. You have the customer, the trust and the address. It just never turned into anything.

Winning a new customer costs advertising, time and negotiating position. Contacting an old one costs one message.

What changes

• The contact happens: on a schedule, not when it occurs to you
• The timing: ahead of the season, while you still have capacity — not at the peak
• The quiet months: spring and autumn fill with predictable work
• Customer data: moves out of the invoices into somewhere usable

A realistic expectation: in the first round only a minority reply — that is normal, and most returning work arrives over the following weeks rather than immediately. From the second year, with history on who responds how, the rate improves noticeably.

What implementation needs

Access to wherever your customer data lives today — the invoicing system, a spreadsheet, or both.

Most of the work is not technical: deciding what counts as due. At what interval, for which equipment, and what to do about customers where you are not sure when you were last there.

Incomplete data is not a blocker. The system works with what exists and flags where it is uncertain — you decide those.

It comes together in two to three weeks, and gets refined from what the first round tells you.

Where it runs

The system connects to your existing invoicing or customer records and runs in your own EU cloud environment.

Your customer list is one of your most valuable assets. It is not used to train a third party's general models, and it does not leave your business.

What it doesn't solve

It does not replace missing data. If all you know about a customer is "Smith, north side", no message comes from that.

It does not persuade someone who does not want maintenance. Some people have not had their air conditioning cleaned in years and will not start now.

And it does not solve a capacity problem. If the messages bring in more work than you can do, that is good news — but then that is the next question, not this one.

Common questions

Won't customers find it intrusive?
An annual service reminder about their own equipment is not advertising. Most customers are grateful, because they genuinely forgot. The difference is in the specifics: "the unit fitted in May 2024" reads differently from "take advantage of our services".

Below how many customers is it not worth it?
Below roughly fifty you can work through them by hand in an afternoon. The system makes a real difference when that no longer fits, or when you have been putting it off for years.

What if I don't have proper records?
Then that is where we start. Invoices usually yield who got what and when — and that has value in itself, regardless of whether you send anything.

Does it go out automatically, or do I see it first?
You see it. Itemised in the early rounds, batched later — but the approval step stays, because you know who you do not want to contact.

How much work will it bring?
There is no honest general figure, because it differs by firm. What is calculable: how many customers you have, how many are due, and what a service is worth. The product is the ceiling, and what materialises is what materialises.

Look at how many customers you last visited two or more years ago. If that number has three digits, you do not need new customers in order to have more work.